Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
Root hairs develop from bulges on root epidermal cells and elongate by tip growth, in which Golgi vesicles are targeted, released and inserted into the plasma membrane on one side of the cell. We studied the role of actin in vesicle delivery and retention by comparing the actin filament configuration during bulge formation, root hair initiation, sustained tip growth, growth termination, and in full-grown hairs. Lipochito-oligosaccharides (LCOs) were used to interfere with growth (De Ruijteret al. 1998,Plant J.13, 341–350), and cytochalasin D (CD) was used to interfere with actin function. Actin filament bundles lie net-axially in cytoplasmic strands in the root hair tube. In the subapex of growing hairs, these bundles flare out into fine bundles. The apex is devoid of actin filament bundles. This subapical actin filament configuration is not present in full-grown hairs; instead, actin filament bundles loop through the tip. After LCO application, the tips of hairs that are terminating growth swell, and a new outgrowth appears from a site in the swelling. At the start of this outgrowth, net-axial fine bundles of actin filaments reappear, and the tip region of the outgrowth is devoid of actin filament bundles. CD at 1.0 μm, which does not affect cytoplasmic streaming, does not inhibit bulge formation and LCO-induced swelling, but inhibits initiation of polar growth from bulges, elongation of root hairs and LCO-induced outgrowth from swellings. We conclude that elongating net-axial fine bundles of actin filaments, which we call FB-actin, function in polar growth by targeting and releasing Golgi vesicles to the vesicle-rich region, while actin filament bundles looping through the tip impede vesicle retention.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Plant cell morphogenesis is cell growth at defined sites in defined directions and amounts. Plant cell growth involves the insertion of Golgi vesicle membranes into the plasma membrane with the simultaneous delivery of Golgi vesicle content into the extracellular matrix, the cell wall. For this exocytosis process, the vesicles have to be transported to the growth site, released and inserted into the plasma membrane. In this study we ask whether the actin cytoskeleton is involved in targeting and releasing the Golgi vesicles to the site of exocytosis, and what the structure of that cytoskeleton is when it supports these functions. We have chosen the root hair for this research since these fast growing cells grow at their tips only, which makes the exocytosis process in the tip of these cells more prominent than in intercalary growing cells.
Growing root hairs have a characteristic pattern of cytoplasmic streaming, in which the cytoplasm flows towards the tip in strands located near the plasma membrane and returns before reaching the tip, often in strands near the center. This pattern of streaming has been called reverse fountain streaming in pollen tubes ( Iwanami 1956). Cytoplasmic streaming in plant cells requires actin filaments. Microinjection of the actin binding protein profilin in Tradescantia stamen hair cells, at concentrations that depolymerized the actin filaments, stopped streaming (mature cells: Staiger et al. 1994 ; growing cells: Valster et al. 1997 ). The streaming pattern in root hairs and other tip-growing cells suggests that this process is somehow involved in targeting of Golgi vesicles from their site of production, the Golgi bodies present in the cytoplasm below the vesicle-rich region, to the base of the vesicle-rich region, where they are released and retained.
Our approach is to compare the configuration of the actin cytoskeleton of root hairs during development. We distinguish four different stages: bulge formation ( Dolan et al. 1994 ) from the epidermal cell, tip-growth of the root hair, growth termination, and full-grown hairs. The study of this developmental process is feasible for root hairs since all developmental stages are present along the root, and from individual hairs the growth status can be determined in the light microscope ( De Ruijter et al. 1998 ;Miller et al. 1997 ). To further probe the relationship between the actin filament configuration and vesicle targeting and release, we applied lipochito-oligosaccharides (LCOs) to reinitiate polar growth ( De Ruijter et al. 1998 ) and used cytochalasin D (CD) to interfere with the actin cytoskeleton.
Because microinjection of fluorescent phalloidin disturbed the polarized cytoarchitecture of vetch root hairs (D.D. Miller, unpublished results), we were not able to study the actin cytoskeleton in living cells, and we therefore used fixed cells. Many electron microscopic studies have shown that freeze-substitution after freeze-fixation is superior to chemical fixation (root hairs: Emons 1987;Emons & Derksen 1986;Galway et al. 1997 ;Ridge 1988). It is also the method of choice for light microscopy ( Baskin et al. 1996 ). Therefore, we have chosen this method for the immuno-localization of actin filaments. We used labeling of sections and whole mounts, of which the sections generally gave the best results. The advantage of sectioning is that no wall-degrading enzymes are needed, which may disturb the cortical actin cytoskeleton. This procedure has the disadvantage that only parts of single hairs can be studied. Since we needed to observe large populations of cells for the drug studies, we improved a chemical fixation procedure so that the cytoarchitecture resembled that of living cells and the actin cytoskeleton resembled that of immuno-labeled cells after freeze-substitution. No wall-degrading enzymes were used and actin filaments were stained with fluorescein-phalloidin.
We show that growth initiation, termination, re-initiation by LCO and termination by CD involve a reorganization of the actin cytoskeleton. These reorganizations are all consistent with two related hypotheses. First, vesicle targeting and release require the presence of net-axially elongating fine bundles of actin filaments (FB-actin). Second, the accumulation and retention of vesicles require the absence of actin filament bundles at the extreme hair tip. We conclude that only when the actin configuration meets both requirements are vesicle targeting, release and retention possible and can tip growth persist.
Root hair growth correlates with elongation of subapical, net-axial fine bundles of actin filaments and the absence of actin filament bundles at the tip
Root hairs are appendages of root epidermal cells, trichoblasts. Before root hair emergence, actin filament bundles in the trichoblasts are oriented mostly longitudinally to the root axis; during root hair development they keep this orientation ( Fig. 1g,h). The first step is the formation of a bulge on a trichoblast ( Dolan et al. 1994 ). In vetch, every epidermal cell forms a bulge and subsequently a root hair (D.D. Miller, unpublished result). The bulge has a fairly triangular shape and consists of a layer of peripheral cytoplasm with strands (s) around a large vacuole (v) ( Fig. 1a). Actin filament bundles in the bulges ( Fig. 1g, arrowhead) are mostly in the same orientation as in the epidermal cell and pass through the cytoplasm at the periphery of the bulge. Bulges develop into growing root hairs which have a smaller diameter, possess polarized cytoplasm ( Fig. 1c1 chemically fixed cell;Fig. 1b,c2,d living cells), and extend in the direction of the tip. The polarized distribution of the cytoplasm is evident in the light microscope by the occurrence of a cytoplasmic dense subapical region ( Fig. 1c1,c2,d; large bracket) that contains organelles and small vacuoles. An even more smooth looking region is located at the apex of the hair ( Fig. 1b,c1,c2,d; small bracket). From electron microscopic studies it is known that this smooth region, of 1–3 μm from the tip, almost exclusively contains secretory vesicles (Vicia sativa, unpublished results; for Vicia villosa, Sherrier & VandenBosch 1994). We will refer to this region as the vesicle-rich region. While the hair grows, the vacuole increases in size but the cytoplasmic dense region, with the vesicle-rich region closest to the tip, remains at the tip ( Fig. 1c).
Actin filament bundles in growing hairs ( Fig. 1g,h,i,j) are present in all the cytoplasmic strands. These bundles lie longitudinally, perpendicular to their net-orientation in the epidermal cell ( Fig. 1g,h). In the subapical region the filament bundles flare out into thinner and thinner bundles and maybe even into single filaments (see also Fig. 4b), but the difference between a single filament of 8 nm and a thin bundle of filaments cannot be resolved in the light microscope. We refer to this subapical actin filament configuration as net-axial fine bundles of actin filaments (FB-actin). In the microscope it can be determined that foci of actin seen in the subapex of growing hairs in micrographs, e.g. Figure 1(h,j,l), are not actin patches but sites where actin filaments bend out of the focal plane. A small region devoid of actin filaments is found at the very apex (arrows in Fig. 1g,h,i,j) and is not yet present in the bulge (arrowhead Fig. 1g). Since actin filaments come nearer to the tip in the cortical than in the central cytoplasm, the region devoid of actin filaments (arrows) is seen as a cleft at the cell apex. This cleft is approximately 5–8 μm wide and 2–6 μm deep ( Fig. 1g,h,i,j), and approximately coincides with the vesicle-rich region. When antibodies are used for actin visualization, a fuzzy staining is seen in the subapex ( Fig. 1i,j), which is not present with fluorescein-phalloidin ( Fig. 1g), indicating the presence of monomeric actin in this region.
The cytoarchitecture of hairs in which growth is terminating, i.e. zone II hairs ( Heidstra et al. 1994 ), is different from that of growing zone I hairs. In zone II hairs the main vacuole is close to the tip, the subapical region with dense cytoplasm is short and contains small vacuoles and numerous cytoplasmic strands that can extend through the tip ( Fig. 1e). Actin filament bundles are found within the cytoplasmic strands in the vacuolated region of these root hairs ( Fig. 1k) and come close to the hair tip, but do not become as thin and densely distributed as in the growing hairs. A region devoid of actin is not present.
Full-grown hairs (zone III hairs) have one large vacuole in the center of the cell and peripherally located cytoplasm ( Fig. 1f1 chemically fixed cell, Fig. 1f2 living cell). The actin filament bundles are found in the layer of cytoplasm at the periphery of the cell also looping through the tip ( Fig. 1l). Not all filament bundles pass through the tip; some turn in the cortical cytoplasm before they reach the apex (s in Fig. 1f2 and arrowhead in Fig. 1m).
In summary ( Fig. 2a), growing hairs possess thick actin filament bundles in the cytoplasmic strands from base to subapex, net-axially aligned fine bundles of actin filaments, FB-actin, in the subapex, and no actin filament bundles in the extreme tip. In hairs that are terminating growth, actin filament bundles come up to the hair tip. In full-grown hairs, thick bundles of actin filaments loop through the extreme tip. We found a correlation between the presence of FB-actin and delivery of vesicles to the base of the vesicle-rich region, as well as between the presence of transverse bundles of actin filaments and the absence of vesicles at the hair tip. Since LCOs reinitiate growth in zone II hairs of vetch ( De Ruijter et al. 1998 ), it is possible to investigate these correlations by applying LCOs to roots.
Lipochito-oligosaccharides reinitiate the elongation of subapical fine bundles of actin filaments and vesicle delivery to a tip region free of actin filament bundles
Vetch root hairs were treated with 10–10 M LCO (Rhizobium leguminosarum bv viciae NodRlv V). After LCO addition the tips of susceptible, zone II root hairs swell ( Fig. 3a) and new outgrowths subsequently develop ( Fig. 3b: arrow, and Fig. 3c–e). In swollen tips ( Fig. 3f2,g3), actin filaments lie close to the plasma membrane ( Fig. 3f1: arrow, and Fig. 3g2) in various orientations ( Fig. 3g1), indicating an absence of polar cytoarchitecture. At some point, the cell changes its growth direction from expansion over the complete surface of the swelling, i.e. undirected growth, to polar growth from one site in the swelling. The initiation of a new site with polar growth ( Fig. 3h3) can be seen (arrow) as a slightly more cytoplasmically dense region. In this region actin accumulates (arrows in Fig. 3h1,h2). As polar growth continues, the cytoarchitecture ( Fig. 3d) and the actin filament configuration in the outgrowth ( Fig. 3i,j) resemble that of growing hairs. Actin filament bundles are again absent from the extreme apex of the new outgrowth, while FB-actin is present subapically (compare Fig. 3i,j with Fig. 1i,j). The schematic illustration in Fig. 2(b) summarizes the results. In freeze-substituted hairs, foci of actin, as reported by Cárdenas et al. (1998) in the outgrowth after LCO-treatment, were not observed. In our study, the actin cytoskeleton in the outgrowth is exactly the same as that in young growing hairs. We have seen such foci in non-optimized chemical fixations and they may be artefacts of the procedure in which high amounts of rhodamine-phalloidin were injected in the Phaseolus root hairs ( Cárdenas et al. 1998 ).
Again, we find that subapical FB-actin in a net-axial orientation together with a zone free of actin filament bundles coincides with tip growth. We test the correlations between the presence of FB-actin and targeting/release of Golgi vesicles, as well as between the presence of transverse bundles of actin filaments in the extreme tip and the absence of Golgi vesicles further by using cytochalasin D.
Cytochalasin D stops elongation of subapical net-axial fine bundles of actin filaments and stops tip growth
Cytochalasin D (CD) binds to the growing end of actin filaments and blocks filament elongation ( Pollard & Mooseker 1981). If growth depends on elongation of FB-actin, CD should terminate root hair growth.
We used CD in concentrations that did not affect cytoplasmic streaming. To prevent recovery in low CD concentrations, we kept the [CD] in the growth medium constant by replacing it for new growth medium with CD every 15 min. No effect was observed in control hairs when PGM without CD was replaced every 15 min. At constant [CD] of 0.5–1.0 μM, cytoplasmic streaming in the hair tube was not affected, but the cytoarchitecture of the tip region in zone I and II hairs changed dramatically. In growing hairs (zone I) the subapical region changed from a region with dense cytoplasm ( Fig. 4a) into a region with cytoplasmic strands dispersed between vacuoles ( Fig. 4c), i.e. a cytoarchitecture similar to that of hairs that stop growing (zone II). At the same time, the hairs that were terminating growth (zone II) obtained the cytoarchitecture of full-grown hairs (zone III) ( Fig. 4e). The effects exerted by CD on root hairs were concentration, duration and developmental stage dependent. Figure 5(a) summarizes the effects of 0.5, 1.0 and 2.0 μM CD on root hair cytoarchitecture, scored after 30 min. Each bar in Fig. 5(a–e) is the mean of the percentages scored on at least 12 plants. Prolonged treatment increased the effects of CD. The percentage of hairs that obtained another cytoarchitecture was doubled when roots were treated for 1 h instead of 30 min with 0.5 μM CD (not shown). Zone II hairs were more sensitive to CD than zone I hairs; in the same time or at lower [CD], more hairs of zone II obtained an altered cytoarchitecture ( Fig. 5a).
Higher concentrations of CD, between 2 and 5 μM, stopped cytoplasmic streaming, but bundles of actin filaments remained present in the cytoplasmic strands (not shown), 10 μM CD killed most root hairs within 15 min.
Actin staining was performed on hairs that had been treated with 1.0 μM CD for 30 min, at which time 75% of these hairs had stopped growing ( Figs 4c and 5a). The subapical region with FB-actin had disappeared in these hairs (compare Fig. 4b,d), and bundles of actin filaments came up to the tip. Thus, CD blocks the elongation of FB-actin and stops growth. This result suggests that elongation of FB-actin is essential for vesicle delivery and confirms our hypothesis that vesicle delivery to the vesicle-rich region requires the presence of such an actin filament configuration. A further test of this hypothesis is the combined use of CD and LCO. Namely, if tip growth indeed depends on the presence of elongating FB-actin, CD should inhibit LCO-initiated root hair outgrowth from swellings.
Cytochalasin D at 1.0 μM does not inhibit swelling and bulge formation, but inhibits polar outgrowth
We applied 10–10 M LCO to roots that were in 0.5 (or 1.0) μM CD ( Fig. 5b,c). After 1 h in LCO, 45% (75%) of zone I hairs had obtained the cytoarchitecture of zone II hairs ( Fig. 4c). Furthermore, their actin configuration had become similar to that of hairs that are terminating growth (zone II) ( Fig. 4d). Irrespective of CD, 90–100% of these hairs started to swell by LCO ( Figs 4g and 5b) in a time course comparable to normal zone II hairs and with a cytoarchitecture similar to swellings in control hairs (compare Fig. 3a with 4g). Furthermore, the actin cytoskeleton in these swellings was similar to that of control swellings which had been incubated in LCO without CD (compare Fig. 3f1 with 4h). Thus, 0.5–1.0 μM CD had no inhibitory effect on LCO-induced swelling of the root hair tip ( Fig. 5b). However, the outgrowth from the swelling was completely inhibited by 0.5 μM CD ( Fig. 5c). These experiments, in which CD and LCO were combined, show that LCO cannot re-initiate tip growth in the presence of CD. This result confirms the hypothesis that vesicle targeting and release to the site of growth require the presence of elongating FB-actin.
In a differently designed experiment we applied LCO first and then treated roots with CD at different stages of the root hair deformation process, and again swelling was not blocked by CD. Furthermore, the initiation of the outgrowth was completely inhibited at concentrations as low as 0.5 μM CD (similar to Fig. 5b,c). When CD was applied at the time LCO had already induced an outgrowth from a swelling, the outgrowth obtained the cytoarchitecture of a hair that stops growing. Thus, this outgrowth behaved as growing hairs in CD, in which the cytoarchitecture changes to that of hairs that are terminating growth, and growth stops prematurely. This experiment corroborates the results obtained when CD was used before LCO.
The cytoarchitecture of the cytoplasm in the swelling resembles the cytoarchitecture of the cytoplasm of the bulges that form on trichoblasts prior to polar growth of root hairs. Both swellings ( Fig. 3a) and bulges ( Fig. 1a) do not have the polar arrangement of the cytoplasm with a localized vesicle-rich region and dense subapical cytoplasm. Furthermore, both stages lack the subapical FB-actin and possess actin filament bundles at the plasma membrane (bulges: Fig. 1g arrowhead, swellings: Fig. 3f1,g1,g2). Therefore, we studied the effect of CD on bulges. Roots growing in 1 μM CD form bulges in high frequencies (85%), while only 15% of these bulges develop into polarly growing root hairs ( Fig. 5d,e). Bulges grown in 1 μM CD ( Fig. 6c) look similar to control bulges ( Fig. 6a). Polarly growing root hairs have a smaller diameter than bulges ( Fig. 6b), but prolonged treatment with CD gives rise to extended bulges, which are as wide as the initial bulges and have no polarized cytoarchitecture ( Fig. 6d). We conclude that in cytoarchitecture and actin cytoskeleton, as well as in sensitivity to CD, the bulges resemble the swellings.
In this paper we report on a specific actin filament configuration in Vicia sativa root hairs during initiation of root hairs on bulges, elongation of root hairs, and growth re-initiation by lipochito-oligosaccharide (LCO) nodulation factors. This typical actin cytoskeleton, which is continuous with bundles of actin filaments in cytoplasmic strands in the basal part of the cell, consists of elongating net-axial fine bundles of actin filaments (FB-actin) in the subapical region, while the cytoplasm in the extreme hair tip is devoid of actin filaments.
Comparison of bulges and swellings
Root hairs with polarized cytoarchitecture originate from bulges, which arise on trichoblasts. The actin cytoskeleton in the bulges does not have the FB-actin and the tip region without actin filament bundles, but has actin filament bundles at the plasma membrane. This actin configuration is also observed in LCO-induced swellings. 1 μM CD does not inhibit bulge formation and swelling, whereas it does inhibit polar growth. This suggests that bulge formation and swelling involve a different mechanism than tip growth.
For bulges and root hairs these observations comply well with the distribution of cytoplasmic calcium ions. Wymer et al. (1997) followed [Ca2+]c prior to and during bulging of the trichoblast of Arabidopsis and found no sustained [Ca2+]c increase preceding the bulging. Furthermore, in the rhd-2 mutant of Arabidopsis, which forms bulges but no root hairs, the elevated [Ca2+]c at the bulge periphery was absent ( Wymer et al. 1997 ). A tip-directed [Ca2+]c gradient was set up in a later stage of bulging and was sustained during root hair growth ( Wymer et al. 1997 ). Such a tip-directed [Ca2+]c gradient at the periphery of the swelling is also set up when an outgrowth forms on the swelling ( De Ruijter et al. 1998 ). Since swelling is an active process which requires mRNA and protein synthesis ( Vijn et al. 1995 ), we think that swelling is growth, but undirected growth.
Bundles of actin filaments at the plasma membrane inhibit exocytosis
In our confocal images of actin skeletons of vetch root hairs, the extreme tip of the hairs is devoid of bundles of actin filaments. This phenomenon has been observed with fluorescence microscopy in several other tip-growing plant and fungal cells that had been prepared by methods other than chemical fixation ( Jackson & Heath 1993;Meske & Hartmann 1995;Miller et al. 1996 ;Roberson 1992). In growing lily pollen tubes injected with fluorescein-phalloidin, a region correlating with the vesicle-rich region is devoid of actin filaments ( Miller et al. 1996 ). Jackson & Heath (1993) observed a cleft devoid of filamentous actin at the tips of growing Saprolegnia hyphae that had been stained with rhodamine-phalloidin during electroporation. When such cells were fixed in formaldehyde, the apical cleft disappeared, leaving the tip diffusely stained. This work confirms that chemical fixation can induce differential actin staining depending on the fixation method employed ( Doris & Steer 1996;He & Wetzstein 1995).
We worked out a chemical fixation procedure that kept the polarized aspect of growing root hairs intact. In this procedure m-maleimidobenzoyl N-hydroxysuccinimide ester was used to stabilize actin filaments prior to fixation by aldehydes ( Sonobe & Shibaoka 1989), and no wall degrading enzymes were used. Furthermore, instead of non-ionic detergents, generally applied to permeabilize and extract cells, we used lysophosphatidyl-choline to facilitate staining by fluorescein-phalloidin without disturbing the fragile cytoplasmic polarity ( Fig. 1c1). By this procedure the vesicle-rich region is devoid of filamentous actin. When fixation was performed without maleimide ester, the polar cytoarchitecture of the hair was often destroyed, and cytoplasmic strands, with actin filament bundles in them, were found all the way to the hair tip.
Braun & Wasteneys (1998) report on an actin patch in the apical dome of characean rhizoids and protonemata after microinjection with rhodamine-phalloidin. The position of the patch coincided with the position of the aggregate of endoplasmatic reticulum in the center of the ‘Spitzenkörper’. They do not report on filamentous actin at the very tip. For root hairs, we cannot conclude that there is absolutely no filamentous actin in the tip region. Short actin filaments, linking cytoskeletal components such as spectrin to the plasma membrane, would probably be missed with the light microscope techniques, and have to be studied in glancing sections with immunogold electron microscopy. However, it is clear that bundles of actin filaments are always absent from the vesicle-rich region, from which we hypothesize that the retention of vesicles at the extreme tip requires an absence of bundles of actin filaments, thus enabling their fusion with the plasma membrane. Also in animal cells during exocytosis, filamentous actin is a barrier to membrane fusion (sperm acrosome reaction: Spungin et al. 1995 ; pancreatic acinar cells: Muallem et al. 1995 ; and chromaffin cells: Tchakarov et al. 1998 ).
Elongating fine bundles of actin filaments deliver Golgi vesicles
If one observes the growing root hair from tip to base, actin filaments are seen first in the region where the transition occurs from vesicle-rich region to subapical region, which contains other organelles besides Golgi vesicles. As far as can be resolved in the light microscope, the region free of actin filaments coincides with the vesicle-rich region. The subapical FB-actin may keep vesicles available for release to the vesicle-rich region at the hair tip. A function as a scaffold, to buffer the interchange between a vesicle reserve and an immediate releasable pool of vesicles, has also been reported for actin filaments in animal cell exocytosis (for review see Yao & Forte 1996). The actin patch ( Czymmek et al. 1996 ) in the ‘Spitzenkörper’ could function in a similar way.
Often, tip growth and intercalary growth of plant cells are seen as two different processes, tip growth being governed by the actin cytoskeleton and intercalary growth by microtubules ( Kropf et al. 1998 ). However, both types of growth rely on exocytosis of Golgi vesicles that have to be brought to the right places for growth to proceed in the right amount and direction. The microtubule cytoskeleton seems to determine the growth direction of a cell ( Baskin et al. 1994 ;Wymer et al. 1996 ), but actin filaments may actually target the vesicles to the plasma membrane. In intercalary growing cells and swelling root hair tips, the presence of FB-actin or single filaments would be difficult to detect with the light microscope, since it can be expected that in these cells the region with Golgi vesicles is not more that one vesicle thick. In this respect, Foissner et al. (1996) obtained interesting data with video microscopy during wound wall formation in characean internodal cells. Wounding of these cells induced a transient reorganization of the actin cytoskeleton from parallel bundles to a fine-meshed network of actin filaments. This fine-meshed network was functional in the movement of Golgi vesicles to the plasma membrane for wall repair. This event is comparable to what we found for root hair tip growth. A similar function for actin filaments, targeting of Golgi vesicles to and retaining them at the plasma membrane, can be envisaged for intercalary growing cells as well.
Cytochalasin D inhibits elongation of FB-actin
CD at 0.5 μM had different effects on the subapical FB-actin than on the thick bundles of actin filaments in the cytoplasmic strands. Actin filament bundles in the cytoplasmic strands remained intact and must have remained functional, since cytoplasmic streaming continued. However, CD transformed the cytoarchitecture of growing hairs into that of full-grown hairs. FB-actin was transformed to thick bundles of actin filaments. The effect of CD on the root hairs can best be explained as an arrest of elongation of the FB-actin at the front of the subapical region, whilst bundling of FB-actin at the base of this region continues. The existing vesicles still incorporate into the plasma membrane at the hair tip, while new vesicles are not delivered to the tip anymore, eventually stopping growth prematurely. This idea also explains the high CD sensitivity of root hairs that terminate growth. Since there are only a few vesicles at the tip of such hairs (zone II), CD alters the type of cytoarchitecture more rapidly and/or by lower [CD] than when there are still many vesicles to incorporate into the plasma membrane (zone I).
Since there are two populations of actin filaments in the growing root hairs, our results shed light on the action of CD in plant cells. It has been reported several times that CD causes bundling of actin filaments in plant cells (pollen tubes: Lancelle & Hepler 1988; root cells: Palevitz 1988; characean internodes: Collings et al. 1995 ). In fact, superficial inspection would have led to the same observation in root hairs. Our results, however, show that CD has the usual depolymerizing effect ( Pollard & Mooseker 1981) on a certain category of actin filaments. The FB-actin at the front of the subapical region of growing hairs disappears. Again, this explanation of the effect of CD clarifies why younger hairs (in zone I) with subapical FB-actin are less sensitive to CD than zone II hairs. Also in the wounded characean internodal cells, CD led to the disappearance of the fine actin filament network, and inhibited transport of vesicles towards the wounded surface ( Foissner & Wasteneys 1997).
Actin filament elongation by LCO occurs after a plasma membrane localized [Ca2+]c increase
The experiments in which LCO was applied to CD-treated roots are useful for understanding the process involved in root hair deformation by LCO. The establishment of polarity, as well as polar growth itself, is inhibited by [CD] as low as 0.5 μM. On the contrary, [CD] even up to 2.0 μM does not block the first morphogenic effect of LCO, i.e. the swelling of the root hair tip. This result implicates that even though swelling is a growth process ( Vijn et al. 1995 ), it does not need new actin polymerization. We know that LCO initiates a membrane localized high [Ca2+]c gradient at the tip of Vicia sativa root hairs ( De Ruijter et al. 1998 ). The [Ca2+]c gradient is present at the plasma membrane of the whole swelling tip. From the result that swelling is independent of actin filament polymerization, it can be deduced that the high [Ca2+]c at the plasma membrane functions in a process that temporally comes before the advent of new actin filament assembly. According to Felle et al. (1998) , working with alfalfa, a Ca2+ influx at the plasma membrane, is the fastest physiological response to Nod factors reported so far. Soon after this, Ca2+ spiking is seen in alfalfa ( Ehrhardt et al. 1996 ). Cárdenas et al. (1998) have shown that in Phaseolus root hairs the actin cytoskeleton disintegrates within 10 min after LCO application, after which it recuperates. It seems that LCO-induced increase of intracellular Ca2+ accompanies actin filament disintegration. We show now that new actin filament elongation is a crucial step for root hair deformation and thus for root hair curling in the presence of bacteria, and that it occurs after calcium ion increase and actin disintegration.
Growth of plant material and root hair deformation
Vicia sativa spp. nigra L. (vetch) seeds were surface sterilized, imbibed in sterile water and placed on 0.8% agarose. Plates were put at 4°C for 3 days to synchronize germination at 20°C. Seedlings were grown in Fåhraeus slides ( Heidstra et al. 1994 ) containing plant growth medium (PGM) consisting of 1.36 m m CaCl2, 0.97 m m MgSO4, 1.12 m m Na2PO4, 1.36 m m KH2PO4 and 20 μm Fe-citrate, pH 6.5 (modified from Fåhraeus 1957). Some seedlings were treated with 10–10 M Rhizobium leguminosarum bv viciae Nodulation (Nod) factors NodRlv V [Ac, C18 : 4] ( Spaink et al. 1991 ) diluted in PGM, while in controls PGM was replaced by PGM.
Optimized ester-aldehyde fixation and fluorescein-phalloidin staining. Root hairs were pre-fixed for 5 min in freshly prepared 200 μm m-maleimido benzoyl N-hydroxysuccinimide ester (MBS, Sigma M2786) in PGM to stabilize actin filaments, fixed with 2% paraformaldehyde and 0.05% glutaraldehyde for 20 min, followed by 4% paraformaldehyde and 0.1% glutaraldehyde for 20 min. Aldehydes were freshly prepared and in Actin Stabilizing Buffer (ASB: 100 m m PIPES pH 6.8, 1 m m MgCl2, 1 m m CaCl2, 75 m m KCl) with 1 m m 4–2-aminoethylbenzene sulfonyl fluoride (AEBSF; Sigma, A8456). Roots were washed three times in ASB buffer prior to permeabilization with 100 μg ml–1 L-α-lysophosphatidylcholine (Sigma, L4129) in ASB. Actin filaments were stained within 10 min with 0.33 μm fluorescein-phalloidin (F-432, Molecular Probes) in ASB with 0.05% acetylated BSA (BSAac) (Aurion, Wageningen, NL) to lower aspecific binding, then mounted in anti-fading agent CITIFLUOR AF3 (Citifluor, Canterbury, UK).
Freeze-substitution and labeling with anti-actin. Roots were removed from the slide; the root tip was excised, placed on a formvar-coated loop, plunge frozen in liquid propane, and substituted for 36–40 h in 3.7% formaldehyde in pure methanol or pure acetone. The sample was brought to room temperature and then rehydrated in series.
For whole mount labeling the cell wall was partially digested with 0.3% cellulase Onozuka R10 (Serva, Heidelberg, Germany), 0.3% cellulysin (Calbiochem, La Jolla, CA, USA) and 0.3% pectinase (Sigma) for 15 min at room temperature. The roots were washed and treated with 1% Triton X-100, 0.05% Nonidet P40, and 2 m m PMSF for 2 min, then washed and blocked with 1% BSAac and 0.05% Tween-20 in PBS for 15 min. Roots were incubated with monoclonal pea anti-actin antibody ( Andersland et al. 1994 ), diluted 1:100, overnight at 4°C. After washing, the secondary antibody, Cy3 conjugated goat anti-mouse Fab fragments (Jackson ImmunoResearch Laboratories Inc.) diluted 1:200, was applied for 2 h followed by washing. Labeled roots were mounted in anti-fading agent and root hairs were imaged with a confocal microscope. Most roots were embedded in butyl methyl methacrylate (BMM) similar to Baskin et al. (1992 , 1996). After substitution the roots were brought to room temperature and washed in acetone or methanol. Infiltration of the resin took place at 2 h intervals with increasing amounts of 80% butyl methacrylate, 20% methyl methacrylate (E. Merck, Darmstadt, Germany), and 0.5% benzoinethylether (Merck) in methanol or acetone; 1:5; 1:2; 1:1; 2:1; 5:1; pure; pure (modified from Baskin et al. 1996 ). The samples were polymerized under long-wavelength UV light at 0°C for 24 h. After embedment, 3 μm sections were prepared. Once sectioned, the BMM was removed by acetone for 25 min followed by three washes in PBS with 0.05% Tween-20. The tissue was blocked with 1% BSAac and 0.05% Tween-20 in PBS for 15 min. Subsequently, the root tissue was incubated in pea anti-actin antibody 1:10 for 2 h at 37°C. Three 10 min washes in PBS with 0.05% Tween-20 followed. Secondary antibody, Cy3 goat anti-mouse antibody, was diluted 1:200 and applied for 2 h at 37°C. Another three 10 min washes in PBS plus 0.05% Tween-20 followed. Labeled sections were mounted in anti-fading agent.
Microscopy and image analysis
Root hairs were observed in the Fåhraeus slides during growth and Nod factor-induced deformation using a Nikon 20× DIC 0.5 NA or 40× Plan DIC 0.7 NA objective on a Nikon Optiphot microscope. Images were recorded on a Panasonic wv-E550 3-CCD camera using a Prysm framegrabber with AcQuis 2.0 software (Synoptics Ltd, Cambridge, UK).
Actin filaments were recorded on a Bio Rad MRC 600 confocal laser scanning microscope (CLSM) equipped with an Argon Krypton laser with a 60× 1.4 NA objective. Neutral density filters were set to obtain 10% transmission intensity from the laser beam using the 488 nm line for fluorescein-phalloidin with DM 488 BA 522 DF 35, and the 568 nm line for Cy3 conjugated goat anti-mouse antibodies with DM 560 long pass BA 585. Optical sections were obtained at 1.0 μm steps with a moderately closed pinhole (setting 3 or 4), using high gain settings and 3–5 Kalman averages. z-Series were projected using Confocal Assistant 2.04. Images were contrast enhanced in Adobe Photoshop (Adobe Systems Inc., Mountain View, CA, USA) and printed with a Kodak XLS 8600 dye sublimation printer.
Cytochalasin treatment of root hairs
The boundaries of the original developmental stages of zones I, II and III of 4-day-old roots were marked on the Fåhraeus slides. Cytochalasin D (CD) (Sigma, C8273) was freshly diluted in PGM from a 20 m m stock in 100% DMSO, to a final concentration of 0.5, 1.0 or 2.0 and applied to the roots, which were incubated in the dark. To prevent recovery of the hairs, [CD] was kept constant by applying fresh CD every 15 min. The percentage of hairs with altered cytoarchitecture was determined every 30 min for 5 h for each [CD]. Each experiment was done on 12 roots or more, and changes in the cytoarchitecture were imaged at the DIC microscope.
To determine the effect of CD on LCO-induced root hair deformation, we first treated roots with 0.5 or 1.0 μM CD for 15 min, then applied 10–10 M LCO NodRlv V [Ac, C18:4] and maintained a constant [CD] for 4 h. The effect of CD on root hair deformation was expressed as percentage of swellings, and percentage of (swellings with) outgrowths. We also applied 10–10 M LCO (t = 0 h) for 15 min, then treated with CD either prior to swellings (t = 1 h), prior to the initiation of an outgrowth from a swelling (t = 1 h 30 min) or during growth of the outgrowth (t = 2 h). The appropriate timing for CD application was determined by a control slide that was incubated with LCO, 15 min prior to the experiment.
We thank Richard Cyr for kindly providing the pea anti-actin antibody, and Allex Haasdijk for the illustration and photographic services. The project was funded by a National Science Foundation Postdoctoral Fellowship in Biosciences Related to the Environment, grant number DBI 9509441, and a Wageningen Agricultural University Postdoctoral Fellowship, both granted to D.D.M.